Claude Fable 5 (xHigh): benchmark results
Provider: Anthropic. Released 2026-06-09. Access: API.
Unified ELO 1781 ± 1, rank #10 of 3081 rated models, from 12 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ARC-AGI-1 | 98.5 | Accuracy (%) | 99.3 |
| Conceptual Reasoning Index - Decision Theory (DTBench) | 95.55 | Chance-Corrected Score (0-100) | 97.1 |
| Conceptual Reasoning Index - Consistency (ACCoRD) | 81.9 | Chance-Corrected Score (0-100) | 96.2 |
| SimpleQA Verified | 70.7 | Accuracy (%) | 96.2 |
| ARC-AGI-2 | 88.33 | Accuracy (%) | 94.4 |
| LiveBench | 78.57 | LiveBench average (self-reported) | 93.6 |
| DuelLab Overall | 60 | DuelLab Score | 93 |
| Conceptual Reasoning Index - Argument Evaluation (LMCA) | 61.02 | Chance-Corrected Score (0-100) | 92.8 |
| ReactBench | 47.1 | Pass@1 (%) | 92.6 |
| Conceptual Reasoning Index | 72.1 | Chance-Corrected Score (0-100) | 92.4 |
| Epoch AI - Cursorbench | 68.4 | Score | 87.7 |
| ObviousBench | 98.61 | Answer pass³ (%) | 86.9 |
Interactive version: theaggregate.ai/model?slug=claude-fable-5-xhigh · How It Works · Data refreshed daily, snapshot 2026-09-21.